New-Onset, But Not Chronic Atrial Fibrillation, Is a Significant Factor Contributing to Mortality Among Patients with Severe COVID-19

Introduction

Atrial fibrillation (AF) is a common rhythm disorder. The frequency of AF can reach up to 0.51% in the general population, with lifetime risk of 25% [1]. AF places a substantial burden on medical care [1]. One large study found that AF onset in the intensive care unit (ICU) was associated with a substantial treatment cost increase ($41,303 vs $28,298, P<0.01), per hospitalization. The difference was even higher when calculations were based among survivors only, with a cost per an ICU hospitalization with AF onset of $81,120, as compared with $59,710 for patients without AF [2]. A study by Fernando et al showed that, in general, the presence of AF onset in the ICU was a significant predictor of total hospital costs (odds ratio [OR] 1.09 [95% CI: 1.02–1.21]) [2]. The higher costs of ICU stays resulted mostly from the length of stay in the ICU and the use of renal replacement therapy, invasive mechanical ventilation, and antiarrhythmic agents [2]. AF influences morbidity and mortality directly or by its complications [3]. Common complications of AF are stroke/embolism, left ventricle dysfunction, heart failure, and cognitive impairment/dementia [3]. AF contributes to emergency department visits, hospitalizations, and lower functional status and quality of life [3], and it is the most common arrhythmia among ICU patients, with a prevalence of 33% [4]. AF observed in the ICU can be present before ICU admission as chronic atrial fibrillation (CAF) or can develop during the ICU stay as new-onset atrial fibrillation (NOAF) [4]. The prevalence of NOAF in the ICU is high, reaching 23% [5]. The high prevalence of AF in the ICU comes also from the fact that patients with CAF are vulnerable to critical illness, due to concomitant diabetes, kidney, and cardiovascular diseases [4]. Furthermore, NOAF can be triggered by critical conditions among individuals admitted to the ICU with sinus rhythm [4]. NOAF episodes in the ICU usually result in further destabilization of a patient’s status. Decreased cardiac output due to NOAF can manifest, for example, in hypotension, shock, heart failure, decreased urine output, acute kidney injury, and altered mental status. It is noteworthy that thromboembolic events triggered by NOAF, mainly stroke, can lead to unfavorable outcomes. The need for anticoagulation due to NOAF can also be challenging in several clinical situations [4]. NOAF during an ICU stay is also a sign of organ damage in critical illness [6,7] and should be perceived similarly to other signs of organ dysfunction, such as impaired consciousness, decreased pO2/FiO2 index, need for inotropic agents, acute kidney injury, and elevated serum lactates [7]. In-hospital morbidity is only the beginning, as NOAF has the tendency do reoccur after hospital discharge, with a 1-year risk of relapse from 16% to even 55%, when a patient develops sepsis during hospitalization [5]. These observations are also true for patients with COVID-19. NOAF and CAF have been investigated as risk factors in this group of patients, with inconsistent results [8–17].

Given that patients with COVID-19 can require ICU admission and that NOAF is common among critically ill patients, we decided to investigate the association between NOAF and several clinical parameters that may serve as potential triggers, as well as the effect of NOAF on patient outcomes. COVID-19 is a disease caused by infection with novel coronavirus SARS-CoV-2 [18]. COVID-19 can manifest with interstitial pneumonia and acute respiratory distress syndrome (ARDS), which is a condition resulting in refractory hypoxemia [19]. ARDS related to COVID-19 is treated primarily with oxygen therapy, high-flow nasal oxygen therapy, and invasive mechanical ventilation [20]. At first, COVID-19 was perceived mostly as a pulmonary disease [19]. Soon, clinicians treating COVID-19 reported complications from the cardiovascular system [21]. COVID-19 is a relatively new phenomenon in medicine, and there is still little research on its prognostic factors. It is known that patients with significant comorbidity are at greater risk of death in the course of SARS-CoV-2 infection. However, there is still no complete data on what medical conditions pose the greatest threat to patients with COVID-19, as well as to what extent the conditions increase the risk of death. The aim of this study was to estimate the odds ratios of death associated with the de novo appearance of AF during disease onset. AF has been identified as a disease with a very high incidence in the ICU, which, at the same time, indicates numerous abnormalities in the body’s homeostasis.

Taking the above into consideration, we decided to retrospectively study the linkage of AF with morbidity and mortality among patients with COVID-19 hospitalized in a single center.

Material and Methods

STUDY DESIGN:

This retrospective, case control study was conducted in the specialist COVID Hospital of Military Institute of Medicine, Warsaw, Poland. We decided to set dichotomous outcomes: in-hospital death and survival. Survival was defined as discharge home or transfer to a ward outside of our COVID-treating hospital. The exposure analysis included 3 groups: the control group, which had sinus rhythm throughout the hospital stay; the NOAF group, for which no information about prior AF was provided by the patient or their relatives, or found in the discharge charts; and the CAF group, for which a history of prior AF was reported by the patient or their relatives, or documented in the discharge charts. Using clinical judgement, we decided to analyze all laboratory tests that were done in most patients that could measure severity of the disease, including markers of inflammation, including ferritin, white blood cells count (WBC), and C-reactive protein (CRP), and markers of global hypoperfusion, including lactates, alanine transaminase, aspartate transaminase, and urea. Based on the literature linking pulmonary hypertension with NOAF, we included the severity of ARDS, as measured by computed tomography (CT) scans, in the analysis [8,22]. All patients admitted consecutively between March 2021 and June 2021 were assessed for eligibility to participate in the study.

STUDY POPULATION:

All patients were adults above the age of 18 years. Women in reproductive age with pregnancy were excluded. As a reference center for severe COVID-19 cases, the hospital had 48 high-dependency beds and 12 ICU beds. All patients were treated by a multidisciplinary team, including an internal medicine specialist, cardiologist, and intensivist. All patients included in this study were admitted to our COVID-19 hospital due to respiratory failure caused by SARS-CoV-2 pneumonia. SARS-CoV-2 infection was confirmed with a polymerase chain reaction test (GeneFinder COVID-19 Plus RealAmp Kit; OSANG Healthcare, Korea). Electrocardiogram (ECG) and medical records were examined for the presence of AF, and a medical interview was obtained from patients and/or from patients’ relatives. All patients with AF were treated according to the existing European Society of Cardiology (ESC) guidelines [3]. This included immediate synchronized DC shock for all unstable arrhythmias and intravenous amiodarone as the first-line treatment for stable patients with NOAF. If amiodarone infusion failed to restore sinus rhythm or to control ventricular rates, patients were administered intravenous metoprolol, or, if patients received a high dose of inotropes, they were administered intravenous digoxin. Patients in stable condition with CAF usually had their long-term therapy prescribed. All patients with NOAF or unstable arrhythmia had their ions corrected. All patients, if indicated, were prescribed a full dose of enoxaparine in 2 doses, adjusted to kidney function. All patients were treated in accordance with the current actual Polish standards at that time, consistent with international COVID-19 treatment guidelines [23]. All treatment strategies for COVID-19, including extracorporeal membrane oxygenation (ECMO), were used as needed. Patients who needed conventional oxygen therapy or high-flow nasal oxygen therapy stayed in the high-dependency unit. Patients who needed mechanical ventilation, vasoactive support, or ECMO were transferred to the ICU. All ICU patients had severe ARDS, neuromuscular blockade, multiple lung recruitment, and prone positioning. All ICU patients needed vasoactive medications, mostly norepinephrine. In the studied sample of 334 patients, 20 patients needed ECMO life support.

MEASURES:

The patients’ ECGs were evaluated with the ESC diagnostic criteria for AF [3]. AF was defined as (1) information about the diagnosis of AF found in medical records from previous hospital or emergency department visits; (2) AF >30 s captured on bedside monitoring; (3) AF documented with 12-lead ECG; or (4) AF resulting in pharmacotherapy or electrotherapy documented in medical records. We did not categorize AF into the subgroups proposed by the ESC, as this was not found to be useful for the purpose of this study [3]. Medical interview and review of medical records was used to identify comorbidities.

The severity of radiological findings was based on the chest CT scale by Francone et al [24]. In this scale, every lung lobe is assessed separately, with 0 points for no changes and 5 points for changes including over 75% of lung parenchyma. The minimal score is 0 points, and the maximal score is 25 points. A total score of 16 or more points indicates very severe ARDS [24]. Laboratory tests and CT scans were performed at admission.

DATA COLLECTION:

The medical records of every patient admitted were reviewed, if suitable data were extracted from the hospital information system. We excluded 29 patients due to missing data or an important medical problem, such as polytrauma, severe traumatic brain injury, cerebrovascular incidents, and out-of-hospital cardiac arrest, ongoing with mild or asymptomatic COVID-19. Finally, we analyzed the charts of 334 patients. A diagram showing the flow of patients is shown in Figure 1.

STATISTICS:

Results were presented as the mean with standard deviation or as median with interquartile range (IQR), according to the normal distribution criteria. Categorical variables were presented as numbers with the occurrence. The Shapiro-Wilk test, which has high power, especially for smaller groups, was used to check the normality of investigated variables. Differences between variables were analyzed with the t test for parametric data, and otherwise with the Mann-Whitney U test. Differences in categorical data were analyzed with the chi-square test. If the number of observations was low, Fisher’s exact test was performed. Univariable and multivariable logistic regression were used to assess associations between dichotomous and continuous data. Only variables significantly differentiating groups and associated with the odds ratios in univariable logistic regression were considered for multivariable analysis. Due to the missing data, multivariable logistic regression analysis could not consider all significantly associated variables simultaneously. Therefore, exclusively for this analysis, missing data from blood sample results were completed with a median, with the abovementioned exception, and raw data were considered for statistics. Two-tailed P<0.05 was considered significant, and P≥0.05 but <0.1 was defined as “trend toward significance”. We used Statistica 13.3 software package (Tibco, USA).

ETHICAL CONSIDERATIONS:

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of Military Institute of Medicine. Authors made an effort to ensure absolute anonymity of used data. Once the database was created, all variables that could possibly jeopardize patients’ anonymity were permanently deleted from the dataset. Datasets were transferred only inside our institution.

Results

Descriptive statistics are presented in Table 1. We performed analysis to identify discrepancies between the groups. The comparison of the NOAF and control groups is presented in Table 2.

The NOAF group had a trend toward significance of being older than the control group and having a higher occurrence of obesity. The sex distribution and hospital stay were not different between the groups. The frequency of comorbidities was similar in terms of hypertension, ischemic heart disease, heart failure, previous myocardial infarction, cerebrovascular disorders, and chronic kidney disease. In both groups, pulmonary embolism was diagnosed with CT with the same frequency during the hospital stay. More patients from the NOAF group had diabetes. Parameters describing the severity of COVID-19 were statistically different in the NOAF group. Inflammation markers CRP, ferritin, and WBC were higher in the NOAF group. Also, the NOAF group had a lower albumin level and higher urea level. Furthermore, severity of lung impairment measured with CT was higher in the NOAF group. Finally, the NOAF group had higher mortality than did the control group.

The comparison between the CAF and control groups is presented in Table 3. The patients in the CAF group were older than those in the control group. The sex distribution and hospital stay were not different between the groups. The CAF group had a higher frequency of comorbidities, with ischemic heart disease, heart failure, previous myocardial infarction, cerebrovascular disorders, and chronic kidney disease. Pulmonary embolism was diagnosed with CT with the same frequency in both groups during the hospital stay. The occurrence of hypertension was similar in both groups. The CAF group had more diabetes, with a trend toward statistical significance. Urea levels were higher in the CAF group than in the control group. Inflammation markers, albumin levels, and CT abnormalities did not differ between the groups. Also, mortality in the CAF group was similar to that of the control group.

Finally, we compared the NOAF and CAF groups. The differences between the NOAF and CAF groups are shown in Table 4. The NOAF population was younger. The sex distribution and hospital stay were not different between the groups. The frequency of comorbidities was similar in terms of obesity, diabetes, hypertension, cerebrovascular diseases, and chronic kidney disease. Ischemic heart disease and heart failure were more frequent in the CAF group. Previous myocardial infarction in the CAF group showed a trend toward being more frequent. The inflammation markers CRP and ferritin were higher in the NOAF group. WBC and urea level were similar. Albumin level showed a statistical trend toward being lower in the NOAF group. The NOAF group had significantly more severe lung involvement on CT. Finally, mortality in the NOAF group was dramatically higher than that in the CAF group. To assess the link between AF occurrence and in-hospital mortality, logistic regression was performed (Table 5). AF was found to be related with the risk of in-hospital death. When the OR was calculated for NOAF and CAF separately, NOAF remained a strong predictor of death, but CAF did not.

To identify factors associated with NOAF, univariable logistic regression analysis was performed. We included variables with substantial differences observed between the NOAF and control groups (Table 2). Then, the variables significantly associated with NOAF were included in the multivariable logistic regression analysis (Table 6). Only severity of CT changes and serum albumin levels were independently associated with the risk of NOAF.

Discussion

LIMITATIONS:

This study had limitations. First, when interpreting the results of this study, it must be considered that the research was designed as a case-control study and had a limited sample size. Second, the study was conducted at a single center and had a retrospective design, with the examined population almost limited to the Polish nationality. Third, in order to include all patients and significantly associated variables in the multivariable regression analysis, we completed missing data with median values, which could influence the achieved results. Therefore, a similar analysis performed on a larger group without missing data could provide more adequate and complete results.

Conclusions

It seems justified to investigate NOAF as sign of organ damage in critically ill patients. Large, multicenter, randomized trials are needed to determine whether NOAF should be perceived as a sign of circulatory instability in systems of risk stratification.

The results of our study, although conducted on just over 300 patients, constitute a significant argument for including AF in the stratification of the risk of death in patients with COVID-19. Given the results, it seems justified to create a risk assessment scale for such patients, which will include the occurrence of a new episode of AF. Also, it seems prudent to test NOAF as a risk factor for death among the general population, for example, in patients with bacterial sepsis.

At the same time, similar relationships can be expected in other serious diseases; however, this requires further research in ICU patient populations with other diseases. In our opinion, it is justified to conduct further studies, preferably prospective ones, on much larger groups of patients with more diverse ethnicities, ages, and comorbidities. Further research would clarify which of the factors and pathomechanisms linking COVID-19 with NOAF are responsible for such a significantly increased mortality in this group of patients.